Finding new materials for future technologies can take years of slow, expensive experiments. However, scientists are now turning to powerful computers and smart data to speed things up. In a recent study, a collaborative team of researchers including Subhasish Mandal, assistant professor in condensed matter physics, used advanced simulations and machine learning to search for rare and promising quantum materials known as altermagnets. These advances open the door to faster breakthroughs in electronics, energy, and beyond.
High-throughput computation enables the rapid generation of data at a scale and pace far beyond what can be achieved experimentally. The materials science community has recently undergone a paradigm shift toward data-driven discovery, where machine learning and computational screening play a central role in discovering novel materials. However, the application of data science to quantum correlated materials has long been constrained by the limitations of existing materials databases.
Mandal and his colleagues from Rutgers University have sought to overcome these barriers by integrating data science with advanced quantum theory to accelerate the discovery of next-generation materials. As a first demonstration, the team employed a high-fidelity high-throughput screening strategy that combines density functional theory (DFT) with embedded dynamical mean-field theory to search for altermagnets.
Altermagnets represent a newly recognized class of magnetic materials, distinct from conventional ferromagnets and antiferromagnets. Their hallmark is a magnetic ordering with vanishing total magnetization yet exhibiting non-relativistic spin-split bands. This unique combination of properties positions altermagnets as very promising for spintronic applications.
Discovering new altermagnets with large splitting through serendipitous search is inherently challenging and resource intensive. The team's high-fidelity high-throughput screening framework substantially improves over conventional DFT-based screening in predicting both metallicity and spin splitting, particularly in transition-metal-rich compounds.
Applying the method developed by the team to more than 2,000 candidate magnetic materials, the researchers:
- identified two previously unreported metallic altermagnets,
- confirmed one known case,
- and discovered over a dozen semiconducting altermagnets.
This work has been published in Physical Review Letters 135, 106501, and has been supported by the National Science Foundation.